Microglial ASD-related genes are involved in oligodendrocyte differentiation

Autism spectrum disorders (ASD) are associated with mutations of chromodomain-helicase DNA-binding protein 8 (Chd8) and tuberous sclerosis complex 2 (Tsc2). Although these ASD-related genes are detected in glial cells such as microglia, the effect of Chd8 or Tsc2 deficiency on microglial functions and microglia-mediated brain development remains unclear. In this study, we investigated the role of microglial Chd8 and Tsc2 in cytokine expression, phagocytosis activity, and neuro/gliogenesis from neural stem cells (NSCs) in vitro. Chd8 or Tsc2 knockdown in microglia reduced insulin-like growth factor-1(Igf1) expression under lipopolysaccharide (LPS) stimulation. In addition, phagocytosis activity was inhibited by Tsc2 deficiency, microglia-mediated oligodendrocyte development was inhibited, in particular, the differentiation of oligodendrocyte precursor cells to oligodendrocytes was prevented by Chd8 or Tsc2 deficiency. These results suggest that ASD-related gene expression in microglia is involved in oligodendrocyte differentiation, which may contribute to the white matter pathology relating to ASD.


Results
Chd8 or Tsc2 was effectively down-regulated by siRNA treatment in primary microglia. To examine the role of ASD-related genes in microglia, we established gene knockdown experiments in primary microglia by siRNA treatment. Initially, we transduced fluorescent-labeled siRNA to primary microglia to determine the efficiency of siRNA transduction and measured the ratio of fluorescent-labeled CD11b + microglia by flow cytometry analysis. The analyses revealed that 94.7% of CD11b + microglia exhibited fluorescence, suggesting that siRNA was effectively transduced to primary microglia (Fig. 1a). The number of microglia was not affected by siRNA treatment, indicating that the treatment was not toxic for primary microglia (Fig. 1b). To down-regulate ASD-related genes in microglia, we transduced control siRNA, Chd8 siRNA or Tsc2 siRNA and examined the expression levels by quantitative RT-PCR after 3 days of transduction. Chd8 siRNA and Tsc2 siRNA effectively down-regulated Chd8 and Tsc2 expression in primary microglia, respectively (Fig. 1c,d). Notably, Gapdh was not affected by Chd8 or Tsc2 siRNA treatment (Fig. 1e). To examine whether Chd8 or Tsc2 deficiency in microglia affects cytokine expression levels, we performed quantitative RT-PCR in microglia after stimulating them with LPS for 6 h 29 . LPS stimulation strongly induced the expression of genes encoding tumor necrosis factor (Tnf), Il1b, Spp1, and reduced by half the expression of Igf1. Although Tnf, Il1b, Spp1 expression were not affected by siRNA treatment, Igf1 expression was reduced in both the Chd8 and Tsc2 siRNA treatment groups compared with the control siRNA treatment group (Fig. 1f-i). These results shows that deficiency in Chd8 or Tsc2 in microglia affects expression of Igf1, but not major pro-inflammatory cytokines.
Knockdown of Tsc2 affected phagocytic activity. Microglia are involved in brain development by engulfing excessive neurons or synapses during brain development 17,30 . To examine whether the deficiency of Chd8 and Tsc2 in microglia affects phagocytosis activity, we performed phagocytosis assays using fluorescentlabeled latex beads. Quantification of the number of fluorescently-labeled beads in Isolectin B4-labeled micro- (c,d) Chd8 (c) and Tsc2 (d) mRNA expression levels were analyzed by quantitative RT-PCR in primary microglia treated with control siRNA, Chd8 siRNA and Tsc2 siRNA. Expression levels of Chd8 and Tsc2 were normalized to that of Gapdh (n = 6, p < 0.001 assessed by the one-way ANOVA, followed by Tukey-Kramer tests). (e) Gapdh expression levels were analyzed by quantitative RT-PCR by normalizing to Actb (n = 4, biologically independent experiments, F (2, 9) = 0.2393, df = 9, p = 0.7920 for main effect of group, p = 0.7789 for control siRNA vs. Chd8 siRNA, p = 0.9012 for control siRNA vs. Tsc2 siRNA assessed by the one-way ANOVA, followed by Tukey-Kramer tests). (f-i) mRNA expression levels of Tnf (f; n = 5, biologically independent experiments, F (3,16)  www.nature.com/scientificreports/ glia indicated a low number of fluorescent-labeled beads in Tsc2 siRNA-treated microglia compared with control ( Fig. 2a,b), whereas microglia treated with Chd8 siRNA did not change. These results indicate that Tsc2, but not Chd8, plays a role in phagocytosis activity in primary microglia.
Deficiency of Chd8 or Tsc2 in microglia led to the impairment of oligodendrocyte development. Microglia promote the differentiation of neurons and oligodendrocytes from NSCs during brain development 15 . Next, we examined whether deficiency of Chd8 or Tsc2 affects the microglia-mediated differentiation of neurons and oligodendrocytes. NSCs were prepared by dissociating neurospheres, and co-cocultured with siRNA-treated microglia across the transwell membranes (Fig. 3a). After 7 days, we counted the number of βIII-Tubulin (Tuj1) + neurons, glial fibrillary acidic protein (GFAP) + astrocytes, platelet-derived growth factor receptor α (PDGFRα) + oligodendrocyte precursor cells (OPC), and myelin basic protein (MBP) + oligodendrocytes by immunocytochemistry. The number of neurons, OPC, and oligodendrocytes, but not astrocytes, significantly increased in the presence of microglia ( Fig. 3b-e) consistent with the previous reports 15 . In addition, the number of MBP + oligodendrocytes significantly decreased in the presence of Chd8 or Tsc2 siRNA-treated microglia compared with control siRNA-treated microglia (Fig. 3e). No change was detected in the number of neurons and OPC in the presence of Chd8 and Tsc2 siRNA-treated microglia (Fig. 3b,d). These results suggest that microglia require Chd8 or Tsc2 to promote the oligodendrocyte differentiation. As OPCs proliferate and differentiate into oligodendrocytes 31 , we asked whether Chd8 or Tsc2 in microglia is involved in OPC proliferation. To this end, we co-cultured OPCs with microglia treated with siRNA and performed immunocytochemistry for Ki67 (proliferation marker), PDGFRα, and oligodendrocyte transcription factor 2 (Olig2) after 3 days (Fig. 3f). We counted the number of Ki67 + proliferating cells in PDGFRα and Olig2 double-positive OPCs, and we found that Chd8 or Tsc2 deficiency in microglia did not affect the percentage of Ki67 + /PDGFRα + Olig2 + proliferating OPCs (Fig. 3g,h). There was no significant difference about total number of Olig2 + cells between the groups (Fig. 3i). To determine whether Chd8 or Tsc2 deficiency in microglia affects oligodendrocyte differentiation, we co-cultured siRNA-treated microglia and NSCs for 7 days and counted the number of MBP + oligodendrocytes to Olig2 + oligodendrocyte-lineage cells. The percentage of MBP + /Olig2 + cells was lower in the presence of Chd8 or Tsc2 siRNA-treated microglia than in control siRNA-treated microglia (Fig. 3j,k). These results suggest that microglia promote the differentiation from OPC to oligodendrocyte in a Chd8-or Tsc2-dependent manner, with no effect on OPC proliferation. Chd8 or Tsc2 knockdown by shRNA in microglia reduced oligodendrocyte differentiation. To assess the sustainable effects of ASD-related genes on microglia, we used lentivirus encoding short hairpin RNA (shRNA) against Chd8 or Tsc2 (Fig. 4a). We confirmed the reduction of Chd8 and Tsc2 expression by quantitative RT-PCR after 7 days of infection (Fig. 4b,c). We then prepared NSCs obtained from neurospheres prepared www.nature.com/scientificreports/ from the P1 subventricular zone (SVZ), where oligodendrocytes are generated 32,33 and developed during the neonatal period. We cultured lentivirus-infected microglia co-cultured with NSCs from the SVZ and counted the number of MBP + oligodendrocytes in DAPI + cells. The number of MBP + oligodendrocytes was reduced in the presence of microglia infected with lentivirus encoding either Chd8 or Tsc2 shRNA compared with microglia infected with scramble shRNA coding lentivirus (Fig. 4d,e). Moreover, we stained with myelin-associated glycoprotein (MAG), which starts to be expressed before MBP in differentiating cultures. The number of MAG + cells was also reduced in the presence of microglia infected with lentivirus encoding either Chd8 or Tsc2 shRNA ( Fig. 4f,g). These results support our finding that Chd8 and Tsc2 play a role in microglia-mediated oligodendrocyte differentiation.
Oligodendrocyte differentiation was also regulated by other neural cells 34 . Therefore, it is possible to suppose that Chd8 or Tsc2 regulates microglia-mediated oligodendrocyte differentiation through other cells. To determine the possibility, we observed the number of neurons, astrocytes, NPCs, and OPC after co-culturing NSCs and microglia infected with lentivirus. The number of βIII-tubulin + neurons or GFAP + astrocytes was not affected by microglia (Fig. 4h-k). The number of Nestin + NPC and Ki67 + cells in Nestin + NPCs was not affected by microglia (Fig. 4l-n). These results suggest that Chd8 or Tsc2 in microglia do not affect NPC proliferation and differentiation of neurons or astrocytes. Although the number of PDGFRα + OPC and the percentage of Ki67 + cells in PDGFRα + OPC was increased in the presence of microglia, Chd8 or Tsc2 knockdown in microglia did not affect the number of OPC, and Ki67 + proliferating OPC compared with microglia infected with scramble shRNA-coding lentivirus ( Fig. 4o-q), suggesting that microglia regulate oligodendrocyte differentiation in Chd8 or Tsc2-dependent manner without affecting neurons, astrocytes, NPC, and OPC.

Chd8 or Tsc2 regulates numerous gene expression in microglia.
To investigate the molecular profiles of microglia with Chd8 or Tsc2 deficiency, we performed RNA sequencing (RNA-seq) in primary microglia treated with control, Chd8, or Tsc2 siRNA after LPS stimulation for three days. Principal component analysis (PCA) showed the difference of transcriptome among control siRNA, Chd8 siRNA, and Tsc2 siRNA treatment microglia (Fig. 5a). We analyzed differentially expressed genes (DEGs) using volcano plots and found that the expression of 83 genes was significantly reduced and 89 genes were increased in Chd8 siRNA-treated microglia ( Fig. 5b,d). Alternatively, the expression of 82 genes was significantly reduced, and 69 genes were increased in Tsc2 siRNA-treated microglia (Fig. 5c,e). To investigate the properties of DEGs, we classified 172 genes (for vs. Chd8 siRNA) and 151 genes (vs. Tsc2 siRNA) by Gene Ontology (GO) analysis of the biological process. Numerous GO terms such as "Cellular localization", "Cellular response to stress", "Protein localization", "Chromatin organization", "Organelle localization", and "Protein modification process" were rich in DEGs of Chd8 siRNA (Fig. 5f). Alternatively, GO terms "Cellular component biogenesis", "Regulation of signal transduction", "Cellular response to stress", "Protein biogenesis", "Demethylation", and "Cell motility" were rich in DEGs of Tsc2 siRNA (Fig. 5g). From these results, both Chd8 and Tsc2 regulate numerous biological processes, especially protein synthesis and localization. We also confirmed the reduction of Tsc2 or Chd8 expression in Tsc2 siRNA-or Chd8 siRNA treated microglia, respectively (Fig. 5h). Moreover, Igf1 expression was also reduced in Chd8 or Tsc2 siRNA treatment (Fig. 5h). These results suggest that either Chd8 or Tsc2 contributes to the expression of genes related to numerous biological processes including Igf1.
Tsc2 deficiency impairs phagocytosis activity in microglia. To determine whether Tsc2 regulates the expression of phagocytosis-related genes, we analyzed the expression of Itgam, Becn1, Itgb2, and Mfge8, which are annotated with phagocytosis engulfment. However, the expression of these genes was not significantly different under Tsc2 siRNA treatment, suggesting that impairment of phagocytosis activity during Tsc2 deficiency was not due to changes in the expression of phagocytosis-related genes (Fig. 5i). Microglia play a role in the engulfment of excessive myelin and synapses 35 . We also analyzed the expression of synapse-or myelin recognition-related genes, such as Msr1, Marco, Cd36, and Trem2. However, there were no significant differences between the groups (Fig. 5j). These results indicate that Tsc2-regulated phagocytosis activity is independent of the expression change of genes that are well-established for phagocytosis recognition.

Chd8 or Tsc2 deficiency in microglia impaired oligodendrogenesis in vivo.
To examine the effect of Chd8 or Tsc2 deficiency in microglia on oligodendrocyte differentiation in vivo, we used AAV6, which has triply mutated capsid variants (mAAV6) 36 to reduce Chd8 or Tsc2 expression in microglia. We inserted enhanced green fluorescent protein (EGFP) and miR-30-based shRNA into the AAV construct under the CD68 promoter to enhance specificity for microglia ( Fig. 6a). At E14, we injected mAAV6 into the embryo lateral ventricle, and performed histological analysis at P12. We confirmed EGFP fluorescence in ionized calcium binding adapter protein 1 (Iba1) + microglia in the corpus callosum (Fig. 6b). Additionally, we purified EGFP + microglia from P12 brain and found that the expression levels of Chd8, Tsc2, and Igf1 in microglia were reduced in Chd8 or Tsc2 shRNA treatment, which was consistent with in vitro studies (Fig. 6c).
To label oligodendrocytes sparsely by immunohistochemistry, we used CC1 monoclonal antibody that recognizes adenomatous polyposis coli (APC). As CC1 monoclonal antibody also recognizes Quaking 7 protein 37 , which is expressed in oligodendrocytes and some astrocytes, we defined oligodendrocytes as CC1 and Sox10 (specific marker of oligodendrocyte-lineage cells) double-positive cells. We counted the number of CC1 + Sox10 + oligodendrocytes and found that a low number of CC1 + Sox10 + oligodendrocytes were detected in the corpus callosum (Fig. 6d,e), anterior commissure (Fig. 6f,g), and striatum (Fig. 6h,i) of mice treated with Chd8 or Tsc2 shRNA compared with control (scramble shRNA treated) mice. To examine the possibility that the reduction of oligodendrocyte number by Chd8 or Tsc2 shRNA was due to the oligodendrocyte death, we observed Cleaved Caspase-3 + Olig2 + double positive cells (apoptotic oligodendrocyte-lineage cells) in the Chd8 or Tsc2 shRNAtreated brain. However, there were few Cleaved Caspase-3 + Olig2 + double positive cells even in Chd8 or Tsc2 www.nature.com/scientificreports/ shRNA-treated brain ( Supplementary Fig. S1a). In addition, there was no significant difference of Olig2 + cell number between the groups ( Supplementary Fig. S1b), indicating that our observation about the reduction of oligodendrocytes does not depend on the difference of oligodendrocyte survival. Alternatively, CC1 + GFAP + cells were not found in these regions (Fig. 6j). GFAP + area (Fig. 6k,l), and the number of PDGFRα + Sox10 + OPC (Fig. 6m,n) was not affected by Chd8 or Tsc2 shRNA treatment in the corpus callosum, anterior commissure, and striatum. These results demonstrate that microglial Chd8 or Tsc2 contribute to oligodendrocyte development in vivo without affecting astrocytes or OPC.

Discussion
Microglia are key players in causing structural abnormalities in the brain during development, and they play a key role in the pathogenesis of neuronal disorders. Regarding the involvement of ASD pathology, microglia engulf synapses of neurons undergoing synapse formation during brain development 17,30 , and this developmental microglial dysfunction causes abnormal behavioral phenotype such as social interaction deficits 38,39 . Considering the presence of ASD-related gene expression in microglia, in this study, we asked whether microglial ASDassociated gene expression is autonomously involved in microglial functions of the cell. According to our qPCR and RNAseq analyses, Igf1 expression was reduced by silencing Chd8 or Tsc2 in primary microglia. During normal development, a subset of microglia located in the white matter promotes myelination by secreting Igf1 16 in a phosphoinositide 3-kinase (PI3K)-Akt-dependent manner 40 . Our cell culture experiments revealed that microglia with Chd8 or Tsc2 knocked down reduced the differentiation efficiency of oligodendrocytes, which is in the same direction as the decrease in Igf1 expression in microglia regulated by ASD-related genes. It has been reported that Igf1 expression is regulated by several transcriptional factors such as CCAAT-enhancer binding protein β (C/EBPβ), C/EBPγ, Interferon regulatory factor 8, and Jun 41,42 . However, our RNAseq analysis could not detect significant difference of these gene expressions in Chd8 or Tsc2 deficient microglia (data not shown). It has also been reported that Chd8 interacts with C/EBPβ and promotes its transactivation activity 43 . In addition, the Tsc1/Tsc2 complex inhibits mTORC1 activity, preventing C/EBPβ translocation 44 . The reduction of Igf1 expression by Chd8 and Tsc2 silencing may be mediated by C/EBPβ inactivation without change of mRNA level of C/EBPβ.
We found that Tsc2 deficiency impaired phagocytosis activity in primary microglia. Tsc2 forms a TSC complex with Tsc1, which inhibits mTOR signaling 10 . Knocking out microglial Tsc1 increases phagocytosis activity 45 . Consistent with this finding, mTOR deletion in microglia led to the impairment of phagocytosis both in vitro and in vivo 46 . In contrast, Tsc2 +/mice showed impaired autophagic activity 47 , and our study showed Tsc2-deficient microglia had reduced phagocytosis activity. This discrepancy may be explained by the difference in signal transduction between Tsc1 and Tsc2. One of the well-established differences between Tsc1 and Tsc2 is the function of Tsc1 in Smad2/3 phosphorylation in the presence of TGF-β, which controls epithelial-to-mesenchymal transition 48 . However, in microglia, TGF-β and Smad3 activation increases phagocytosis activity 49 ; therefore, TSC-mediated intracellular signaling may have cell-type specific regulation. Further studies on the intracellular mechanism will contribute to the understanding of Tsc2-mediated intracellular signaling that regulates phagocytosis.
In this study, we found a decrease in the number of oligodendrocytes generated from NSCs or OPCs cocultured with Chd8-or Tsc2-deficient microglia. However, we did detect that Tsc2-deficient microglia showed altered phagocytosis activity. As we used transwells for co-culture experiments, the decrease in the number of  www.nature.com/scientificreports/ oligodendrocytes that we observed does not depend on the phagocytic function of microglia in vitro. However, we should note that the recent study showed that microglia regulate myelination by engulfment of OPC in developing brain 19 . It is generally accepted that microglia eliminate synapses and myelin during brain development in a cell-cell contact-dependent manner. Molecules involved in phagocytosis of synapses and myelin by microglia included complement receptor-3 (Itgam), scavenger receptor (Msr1, Marco, Cd36), and Trem2 50 . Although our RNAseq analysis showed that Chd8 or Tsc2 deficiency did not affect the expression of the well-established genes which is associated with the phagocytosis of synapses and myelin, there is still the possibility that ASD-related microglia contribute to phagocytosis of synapses and myelin because the expression of some of the above molecules is dramatically regulated by several conditions (such as developmental regulation of Trem2) 51 .
Previous studies have demonstrated that myelin dysfunction was observed in ASD patients 4 , and genetic analysis showed that myelin development-related genes were associated with ASD 52,53 . Therefore, dysfunctional myelin development may be key to understanding ASD pathology. Regarding Chd8 and Tsc2, oligodendrocytespecific loss of function experiments showed the impairment of oligodendrocyte differentiation and causes hypomyelination through the epigenetic modulation of gene expression, which is involved in oligodendrocyte proliferation and myelination 8,9,13 . In this study, we focused on the expression of ASD-related genes in microglia and revealed a novel function of ASD-related genes in myelin pathology. However, we should note that the expression of ASD-related genes is not limited to the brain, but is also detected systemically. Recently, we found that changes in systemic environments control myelin development directly via circulating factors 54 , providing more opportunities to understand the contribution of ASD-related genes to neuropathology.

Materials and methods
Animals. Pregnant mice at embryonic day (E) 14 and E18 of C57BL/6J mice were purchased from SLC Japan.
All experiments procedures were approved by the Animal Care Committee of National Center of Neurology and Psychiatry (2018034R9). The mice were housed in an air-conditioned room at 23 ± 1 °C with a 12-h light-dark cycle under specific pathogen-free conditions and had free access to water and food. The mice were randomly assigned to groups. All experiments were conducted according to the relevant guidelines and regulations including Animal Research Reporting of In Vivo Experiments (ARRIVE) guidelines.
Microglia and OPC primary culture. Whole brains were dissected from E18 mice, and incubated with 0.25% trypsin (Gibco) in Dulbecco's Modified Eagle Medium (DMEM; Gibco) for 15 min at 37 °C, followed by treatment with DNase I (Sigma Aldrich) for 1 min at 37 °C. Cell suspension was washed by DMEM containing 10% fetal bovine serum (FBS) and centrifuged at 450×g for 10 min. The isolated cells were resuspended with www.nature.com/scientificreports/ DMEM containing 10% FBS, and filtered with 70 μm nylon cell strainer. Cells were plated on poly-l-lysine precoated 75 cm 2 tissue culture flask (IWAKI) in DMEM supplemented with 10% FBS and penicillin-streptomycin (Thermo Fisher Scientific). After 12-15 days, culture flasks were gently shaking for 30 min, and supernatant was centrifuged at 1500 rpm for 10 min. The cell pellet was resuspended with microglia culture medium supplemented with 50% of the supernatant, 10% FBS in DMEM, and plated to 96 well plate. Each biological replicate represents a culture prepared from a distinct mouse brain. All experiments were repeated at least three times. For primary OPCs, mixed glial cells with removed microglia were detached from the culture flask by treatment with 0.05% trypsin for 5 min and replated to fresh culture flasks. After 30 min of incubation, the culture medium was centrifuged at 500×g, and the cell pellet was resuspended in OPC culture medium supplemented with sodium pyruvate (1:100, Gibco), apo-transferrin (50 μg/mL, Sigma-Aldrich), bovine serum albumin (0.1%, Sigma-Aldrich), insulin (5 μg/mL, Sigma-Aldrich), biotin (10 nM, Sigma-Aldrich), hydrocortisone (10 nM, Sigma-Aldrich), sodium selenite (30 nM, Sigma-Aldrich), and penicillin-streptomycin (1:100, Life Technologies). OPCs were plated onto poly-l-lysine-coated 96 well plates.   36 . pAAV, pHelper, and modified AAV6 capsids were transfected to AAV293 cells (Takara) by PEI Max. After 3 days, AAV was purified with an AAVpro Purification kit (Takara) and concentrated with an Amicon Ultra-15 (Millipore). To inject AAV into the mouse brain, E14 pregnant mice were anesthetized with isoflurane (3%), and 1 μL of AAV containing 0.01% fast green was injected into the lateral ventricle of the embryonic mouse brain, and P12 neonatal mice were analyzed.